CHAPTER 100 Epilepsies 1649

of language cortex requires the patient to be involved in language
activities during electrical stimulation. Depth electrodes implanted
stereotactically have become the dominant method of invasive EEG
recording due to greater tolerability, lower complication rate, and
superior flexibility (Mullin et al., 2016; Schmidt et al., 2016).
MEDICAL THERAPY
Medical therapy is generally the first-line treatment after the diagno-
sis of epilepsy has been made (Fig. 100.19). Its goal should be com-
plete seizure freedom in the absence of medication side effects. Since
the choice of therapy depends on seizure and epilepsy classification,
ideally there should be enough evidence to diagnose the seizure type
and the epilepsy syndrome prior to deciding on initial therapy. The
treating physician must have a low threshold for reevaluating the
diagnosis when therapy fails, particularly when the diagnosis of epi-
lepsy is based on history alone without supportive EEG or imaging
data.
Pharmacotherapy is not always necessary after a single unprovoked
seizure, particularly when the risk of recurrence appears low, based
on known predictors (e.g., with a normal EEG and MRI; see earlier
discussion of epidemiology of the first unprovoked seizure) (Krumholz
et al., 2015). Delay in initiation of therapy until after the second unpro-
voked seizure does not affect the odds of subsequent long-term remis-
sion. Rarely, some mild forms of epilepsy may not require therapy.
For example, in BECTS, seizures may be infrequent, occur in sleep,
and remit at puberty. In some patients with JME, seizures are clearly
linked to sleep deprivation or binge alcohol consumption, and may be
managed with lifestyle adjustments only, especially when myoclonic
seizures are the only seizure type. In the vast majority of individuals,
however, medical therapy is necessary.
Initiating Therapy
The many ASMs available for prescription include classical ASMs that
were marketed before 1980 and many new ASMs that have been mar-
keted since 1993. Knowledge of their pharmacokinetic properties is
essential for safe and effective use (Table 100.2). Choice of the first ASM
depends primarily on the classification of seizure type and epilepsy syn-
drome (Table 100.3) but should also take into consideration factors such
as age, gender, comorbid conditions, and individual circumstances. The
ASM choice should consider the ASM’s spectrum of efficacy, its specific
pharmacological properties in relation to the patient’s specific needs, its

Epilepsy diagnosis

First monotherapy

Second monotherapy or
adjunctive therapy

Refer to epilepsy center

Reassess diagnosis; video-EEG Not epilepsy
monitoring Revise therapy

Structural
Consider/evaluate for epilepsy
Nonstructural with Yes, proceed with
surgery for TLE with
generalized onset seizures presurgical evaluation,
hippocampal sclerosis or focal
then epilepsy surgery
epilepsy with well-defined and
resectable epileptogenic lesion

Revise ASM therapy

Continue trials of medical No Dietary therapy may be
therapy: additional one to Continue trials of medical considered sooner in
five ASM regimens therapy: additional one to five motivated individuals or
ASM regimens those that are tube-fed

Consider VNS, dietary In event of persistent ASM

therapy, neurostimulation,
corpus callosotomy
failure (whether due to lack of
Good candidate—
efficacy or poor tolerability), re-
proceed with epilepsy
evaluate for epilepsy surgery,
surgery
including invasive EEG-video
monitoring if appropriate

Not a surgical candidate

Consider VNS, dietary therapy,
neurostimulation, palliative
epilepsy surgery

Fig. 100.19 Treatment Algorithm for Epilepsy. Additional steps assume that seizures are not controlled
despite adequate trial of well-tolerated medication. AED, Antiepileptic drug; EEG, electroencephalographic;
TLE, temporal lobe epilepsy; VNS, vagus nerve stimulation.

F ECF FH HE @ @AE C @AE C HE @ D AC FD 02 .4.(1( 4 ( 1. 4. (0 4). 1) 4 ) 1 HFD

1650 PART III Neurological Diseases and Their Treatment

safety profile in the patient’s age and gender group, as well as its efficacy
in the patient’s comorbid conditions. Whenever possible, these determi-
nations should be based on solid evidence derived from well-designed
studies. The ASMs with FDA-approved indications outside of epilepsy
include clonazepam (panic attacks); carbamazepine (trigeminal neural-
gia, bipolar disorder); valproate (migraine prophylaxis, acute treatment,
and maintenance for mania/bipolar disorder); gabapentin (postherpetic
neuralgia); lamotrigine (maintenance for bipolar disorder); topiramate
(migraine prophylaxis); and pregabalin (diabetic peripheral neuropathy,
postherpetic neuralgia, fibromyalgia). However, several ASMs are used
off-label for a variety of nonepilepsy indications ranging from insom-
nia to restless legs syndrome and essential tremor, based on trial data or
anecdotal evidence (Azar and Abou-Khalil, 2008).
For focal-onset seizures, carbamazepine or phenytoin may be used
first, but newer drugs have clear pharmacokinetic advantages, particu-
larly absence of enzyme induction (French et al., 2004). Oxcarbazepine
and topiramate were the only ones with official FDA indications, until
2016 when the FDA allowed a drug’s efficacy as adjunctive therapy in
adults to be extrapolated to efficacy in monotherapy (Abou-Khalil,
2019). Lamotrigine, gabapentin, levetiracetam, zonisamide, lacosamide,
and eslicarbazepine acetate have clinical trial evidence supporting their
use as initial monotherapy (Kanner et al., 2018a). A large community-
based study found that lamotrigine was significantly better than carba-
mazepine, gabapentin, and topiramate and had a nonsignificant advan-
tage compared to oxcarbazepine with respect to time to treatment failure
(Marson et al., 2007a). However, lamotrigine requires slow titration and
would not be an appropriate first choice when a rapid onset of action is
needed. When rapid therapeutic effect is required, oxcarbazepine and
levetiracetam may be the drugs of choice because they can be started at
an effective dose. Topiramate also requires slow titration. Because of its
cognitive adverse effects, it is not generally the first drug of choice unless
comorbidities (e.g., migraine, obesity) favor its use.
For generalized-onset seizures, the initial ASM is dependent on the
seizure type(s). For pure generalized absence seizures, ethosuximide
is the first drug of choice, based on a comparative trial with valproate
and lamotrigine, in which it had the best balance of efficacy and toler-
ability (Glauser et al., 2010). Valproate was equally effective and may
be the best choice if there are concomitant GTC seizures or generalized
myoclonic seizures because ethosuximide efficacy is limited to gen-
eralized absence seizures. For IGE with GTC seizures, valproate was

TABLE 100.2 Antiepileptic Drug Absorption, Elimination Half-Life, Formulations (Displayed in

Order They Were Marketed in the United States)
Oral Bioavailability Half-Life (Hours)*

High: ≥90% Short: ≤10

Antiseizure Intermediate: ≥70–<90% Intermediate: >10–<30 Intravenous Extended-Release Requires Slow
Medication Low: <70% Long: ≥30 Formulation Oral Formulation Titration
Phenobarbital High Long X
Primidone High Intermediate X
Phenytoin High Intermediate X† X
Methsuximide‡ High Long
Ethosuximide High Long
Clonazepam High Long
Carbamazepine Intermediate Intermediate X X
Valproate High Intermediate X X
Felbamate High Intermediate
Gabapentin Low§ Short
Lamotrigine High Intermediate X X
Topiramate Intermediate Intermediate X X
Tiagabine High Short X
Levetiracetam High Short X X
Oxcarbazepine‡ High Intermediate X
Zonisamide High Long
Pregabalin High Short
Lacosamide High Intermediate X
Rufinamide Intermediate Short X
Vigabatrin High Short¶
Ezogabine/retigabine Low Short X
Clobazam High Long
Perampanel High Long
Eslicarbazepine High Intermediate
Brivaracetam High Short X
Cannabidiol Low Long +
Cenobamate Intermediate Long X
*The t1/2 for antiseizure medication (ASM) given as monotherapy in a young adult.
†Parenteral formulation is also available as fosphenytoin, a phenytoin prodrug that can be administered intramuscularly.
‡Applies to active metabolites (N-desmethyl methsuximide for methsuximide, monohydroxy derivative [MHD] for oxcarbazepine).
§Gabapentin bioavailability decreases with increasing dose, ranging from ≈60% after a single 300 mg dose to ≈30% at 4800 mg/day given in three

divided doses.
¶The duration of ASM effect is much longer than expected for the short t .
1/2

F ECF FH HE @ @AE C @AE C HE @ D AC FD 02 .4.(1( 4 ( 1. 4. (0 4). 1) 4 ) 1 HFD

 CHAPTER 100 Epilepsies 1651

TABLE 100.3 Established Efficacy of Antiseizure Medications by Seizure Type (FDA Indications
and Class I–III Evidence)
Fda-Approved Indications
Other
Monotherapy Versus Noteworthy
Adjunctive Therapy Efficacy
Indication* Seizure Type or Syndrome Indication Evidence

Antiseizure Generalized Therapy and

Medication Monotherapy Adjunctive Focal Tonic-Clonic Absence Myoclonic LGS IS Seizure Type
Phenobarbital X X X
Primidone X X X
Phenytoin X X X X
Methsuximide X X X Two class IV trials
supporting efficacy
for focal-onset
seizures
Ethosuximide X X X
Clonazepam X X X† X
Carbamaze- X X X
pine
Valproate X‡ X X X‡ X X‡ Initial monotherapy
for generalized
tonic-clonic and
myoclonic seizures
Felbamate Conversion to X X X‖ X‖
monotherapy§
Gabapentin X X Initial monotherapy
for focal seizures
Lamotrigine Conversion to X X X X‖ Initial monother-
monotherapy§ apy for absence
seizures
Topiramate X X X X X‖
Tiagabine X X
Levetiracetam X X X X Initial monotherapy
for focal seizures
Oxcarbazepine X X X
Zonisamide X X Initial monotherapy
for focal seizures
Pregabalin X X
Lacosamide X X X
Rufinamide X X‖
Vigabatrin X¶ X X X¶
Ezogabine/ X X
retigabine
Clobazam X X
Perampanel X X X X Efficacy against
myoclonic seizures
Eslicarbaze- X X X
pine
Brivaracetam X X X
Cannabidiol X X (also DS)
Cenobamate X X X
DS, Dravet syndrome; FDA, US Food and Drug Administration; IS, infantile spasms; LGS, Lennox-Gastaut syndrome.
*The FDA indications for older antiseizure medications (ASMs) (first 7 rows) are less specific with respect to monotherapy versus adjunctive indication,
and even with respect to seizure type. Official labeling may not specify monotherapy versus adjunctive therapy for older ASMs marketed prior to 1993.
†For patients who have failed ethosuximide.
‡The official FDA indication is initial monotherapy for absence or partial seizures and adjunctive therapy “in patients with multiple seizure types that

include absence seizures.” Valproate is widely considered a monotherapy drug of choice in patients with idiopathic generalized epilepsy with tonic-
clonic or myoclonic seizures.
§For partial-onset seizures only.
‖Adjunctive in LGS.
¶Monotherapy only for infantile spasm.

F ECF FH HE @ @AE C @AE C HE @ D AC FD 02 .4.(1( 4 ( 1. 4. (0 4). 1) 4 ) 1 HFD

1652 PART III Neurological Diseases and Their Treatment
significantly better than both lamotrigine and topiramate for time to
treatment failure (Marson et al., 2007b) and may be the first drug of
choice for men, in the absence of prohibitive comorbidities. However,
valproate is teratogenic, with dose-related increased risk of major con-
genital malformation, permanent cognitive impairment, and increased
risk of autism in the exposed fetus. A different ASM should be used
first in women with IGE, but valproate may be used at a low dose if
other ASMs fail to control seizures (Montouris and Abou-Khalil,
2009). While no ASM has official FDA initial monotherapy indica-
tion for generalized myoclonic seizures, valproate is clearly effective,
and levetiracetam, which has FDA approval as adjunctive therapy for
generalized myoclonic seizures (Noachtar et al., 2008), may also be
effective in monotherapy. Weaker evidence exists for efficacy of topi-
ramate, zonisamide, and lamotrigine (lamotrigine may even aggravate
myoclonic seizures in some individuals) (Biton and Bourgeois, 2005;
Crespel et al., 2005; Genton et al., 2006; Prasad et al., 2003). The newest
drugs perampanel and brivaracetam also have anecdotal evidence of
efficacy.
For all epilepsy indications, treatment is initiated with an ASM
monotherapy. In the absence of urgency, it is preferable to start at a low
dose and titrate slowly, even for ASMs that can be started at a higher
effective dose. The initial target dose is often the minimal effective dose
that has been demonstrated in clinical trials, keeping in mind that the
pivotal clinical trials may have underestimated or overestimated that
dose in some instances. If the initial target dose is not sufficient, the
ASM dose can then be titrated gradually until efficacy is established.
For patients with infrequent seizures, it may take a long time to deter-
mine when an effective dose has been reached. Therefore it is wise for
the initial target dose to be an average rather than a minimum effective
dose. Before a medication can be considered ineffective, it usually has
to be titrated to the highest tolerated dose. If a medication fails due to
lack of efficacy, the neurologist may choose either replacement mono-
therapy or adjunctive therapy with another medication. The available
evidence is that the two options do not differ significantly in either
efficacy or tolerability (Beghi et al., 2003; Kwan and Brodie, 2000b). If
the initial therapy has been completely ineffective, then replacement
monotherapy is the best choice. If the initial therapy was partially effec-
tive, adjunctive therapy may be a consideration. If medication failure is
due to lack of tolerability, then replacement monotherapy is the clearly
preferable option. Replacement monotherapy usually requires initially
adding the new ASM before withdrawing the old agent. However,
overnight switch is possible for some ASMs such as carbamazepine and
oxcarbazepine (Albani et al., 2004).
Adjunctive therapy should take into consideration any possible phar-
macodynamic or pharmacokinetic interactions between the medications
in question (Table 100.4). Ideally, the added medication should not have
adverse pharmacokinetic or pharmacodynamic interactions (Abou-
Khalil, 2017). All ASMs are appropriate for adjunctive therapy (Kanner
et al., 2018b). Some ASM combinations have a suggestion of synergy.
For example, the lamotrigine and valproate combination seems to have
greater efficacy than what would be predicted by the efficacy of each
ASM alone (Brodie and Yuen, 1997). There is also evidence favoring the
combination of lamotrigine and levetiracetam (Kinirons et al., 2006).
At present, the ASM mechanism of action (Table 100.5) is not cru-
cial for initial ASM selection (Guimaraes and Ribeiro, 2010), but there
is a suggestion that combining two ASMs with different mechanisms
may have a greater chance of efficacy than combining two ASMs with
the same mechanism (Brodie and Sills, 2011; Margolis et al., 2014).
Mechanism of action may be a predictor of adverse effects from phar-
macodynamic interactions. For example, dizziness, ataxia, and diplo-
pia are more likely when combining lacosamide with another agent
that acts on the sodium channel (Novy et al., 2011; Sake et al., 2010).
F ECF FH HE @ @AE C @AE C HE @ D AC FD
The neurologist will often need to reduce the dose of the initial ASM
when adding a second ASM with a similar mechanism of action. A
better understanding of the epilepsy pathophysiology in individ-
ual patients may improve the selection of ASMs in the future, with a
greater role for mechanism of action.
There are limited examples of epilepsy genetics predicting ASM
efficacy. Autosomal dominant nocturnal frontal lobe epilepsy is par-
ticularly responsive to carbamazepine and its analog, oxcarbazepine
(Picard et al., 1999; Raju et al., 2007). Dravet syndrome, which is usu-
ally caused by a sodium channel mutation, may have seizure aggrava-
tion with the use of agents acting on the sodium channel, particularly
lamotrigine (Guerrini et al., 1998).
Antiseizure Medication Considerations Based on Age
and Gender
Age and gender may have an influence on ASM selection (Azar
and Abou-Khalil, 2008). In the pediatric age group, specific epi-
lepsy syndromes have implications for ASM efficacy. Tolerability
profiles may also be different for children and adults, which may
influence the risk/benefit ratio for specific agents. Serious rashes
from lamotrigine, behavioral adverse effects from levetiracetam,
and oligohidrosis from topiramate and zonisamide are more likely
in children, while hyponatremia from oxcarbazepine and aplastic
anemia from felbamate are much less likely. Valproate-induced
liver failure is more likely in children younger than 2 years of age.
In women of childbearing potential, certain ASMs may reduce the
efficacy of oral contraception, and valproate is to be avoided for
two important reasons: higher risk of congenital malformations
in the exposed fetus and increased risk of polycystic ovaries and
hyperandrogenism (Lofgren et al., 2007).
In the elderly, a key consideration is interaction with other medi-
cations, which favors newer nonenzyme-inducing ASMs (Arain et al.,
2009). Some ASM adverse effects are more likely in the elderly—for
example, reversible parkinsonism and cognitive impairment from val-
proate (Armon et al., 1996) and hyponatremia from oxcarbazepine,
particularly when combined with diuretics or other agents that may
lower sodium (Dong et al., 2005). Several trials have examined the com-
parative efficacy and tolerability of ASMs in the elderly. Lamotrigine
and gabapentin were better tolerated than immediate-release carba-
mazepine in new-onset geriatric epilepsy (Brodie et al., 1999; Rowan
et al., 2005), but no difference was found between lamotrigine and
extended-release carbamazepine (Saetre et al., 2007).
Pharmacoresistance, Tolerance, and Seizure Aggrava-
tion
Persistence of seizures despite ASM therapy should always prompt
reevaluation of the diagnosis; one of the most common causes of
apparent drug resistance is misdiagnosis of nonepileptic psycho-
genic seizures. Other causes of apparent pharmacoresistance include
breakthrough seizures related to noncompliance, sleep deprivation
(particularly in IGE), alcohol or drug abuse, and co-medications that
reduce the seizure threshold. In addition, seizures may appear resistant
because of incorrect ASM selection or inadequate ASM use.
Approximately a third of individuals with epilepsy will not become
seizure free at follow-up despite adequate ASM therapy (Chen et al.,
2018; Kwan and Brodie, 2000a). However, the definition of the exact
time point when epilepsy is considered drug resistant has not been
agreed upon. An ILAE task force recommended the definition of
drug-resistant epilepsy as “failure of adequate trials of two tolerated,
appropriately chosen and used antiepileptic drug schedules (whether
as monotherapies or in combination) to achieve sustained seizure free-
dom” (Kwan et al., 2010).
02 .4.(1( 4 ( 1. 4. (0 4). 1) 4 ) 1 HFD
 CHAPTER 100 Epilepsies 1653

TABLE 100.4 Hepatic Metabolism, Enzyme Induction/Inhibition, Pharmacokinetic Interactions,

and Protein Binding
Hepatic Affected Affected
Hepatic Hepatic Enzyme Enzyme by Enzyme by Enzyme Protein
Metabolism Induction Autoinduction Inhibition Inducers Inhibitors Binding

High: >90% +Minimal

Intermediate: ++Intermediate
Antiseizure ≥50% to ≤90% +++Pronounced +Affected High: ≥85%
Medication Low: <50% −Absent −Not Affected Low: <85%
Phenobarbital Intermediate +++ − − + + Low
Primidone Intermediate +++ − − + + Low
Phenytoin High +++ − − + + High
Methsuximide Low + − − + − Low
Ethosuximide Intermediate − − − + − Low
Clonazepam Intermediate − − − − − High
Carbamazepine High +++ +++ − + + Low
Valproate High − − +++ + + High
Felbamate Intermediate + − ++ + + Low
Gabapentin None − − − − − None
Lamotrigine High + + − + + None
Topiramate Low +* − +* + − Low
Tiagabine High − − − + − High
Levetiracetam None − − − ± − Low
Oxcarbazepine High ++† − +† + + Low
Zonisamide Intermediate − − − + + Low
Pregabalin None − − − − − None
Lacosamide Low − − − + − Low
Rufinamide High + − + + + Low
Vigabatrin None + − − − − None
Ezogabine/reti- Intermediate − − − + − Low
gabine
Clobazam High + − + + + Low
Perampanel High + − − + − High
Eslicarbazepine High + − + + − Low
Brivaracetam High − − − + − Low
Cannabidiol High − − + + + High
Cenobamate High + − ++ + − Low
*Applies to dose ≥200 mg.
†Applies to dose ≥900 mg.

Drug resistance may be related to the underlying epilepsy pathol-
ogy. Idiopathic epilepsy is less likely to be drug resistant than symp-
tomatic epilepsy (Berg et al., 2001b). In one large study, 82% of
patients with generalized idiopathic epilepsy were seizure free for the
past year as compared with only 35% of symptomatic focal epilepsy,
25% of patients with cerebral dysgenesis, 11% of patients with tempo-
ral lobe epilepsy due to mesial temporal sclerosis, and 3% of patients
with dual pathology (Semah et al., 1998). Initial seizure frequency was
also a predictor (Berg et al., 2001a). In one study, drug resistance was
seen in 51% of individuals who had more than 20 seizures before start-
ing treatment, versus 29% of those who had fewer than 20 seizures
(Kwan and Brodie, 2000a). Other predictors of drug resistance were
history of acute symptomatic or neonatal status epilepticus and focal
EEG slow activity. Age at onset between 5 and 9 years predicted a lower
risk of resistance (Berg et al., 2001a).
Resistance to the first ASM predicts resistance to the next ASM. In
the landmark study of Kwan and Brodie, 47% of patients with newly
treated epilepsy became seizure free with the first ASM monotherapy,
13% with the second ASM monotherapy, and 1% with the third ASM
monotherapy. Among patients who failed the first ASM, 11% became
seizure free if the first ASM was ineffective, while 41%–55% became
seizure free if the first ASM was not tolerated owing to side effects or
idiosyncratic reaction (Kwan and Brodie, 2000a). In a follow-up report
of 1795 patients, the 1-year seizure-free rate was 50.5% on the first
drug regimen, 11.6% on the second regimen, and 4.4% on the third
regimen (Chen et al., 2018).
Other studies have suggested that continued ASM trials have a
greater chance of achieving remission than suggested by the study
of Kwan and Brodie. In one study of continued ASM trials in drug-
resistant epilepsy, 14% achieved a 6-month terminal seizure remission
after 3 years of follow-up with medication therapy only (Callaghan
et al., 2007). In a separate study, about 16% of all drug introductions
resulted in seizure freedom for more than 12 months, and 28% of
patients were rendered seizure free by a drug introduction over 5 years
of follow-up (Luciano and Shorvon, 2007). Negative clinical predictors
in these studies included history of status epilepticus, younger age at
intractability, greater number of failed drug therapies, and presence
of intellectual disability. Shorter-duration epilepsy and idiopathic epi-
lepsy were favorable predictors. Failure of ASM therapy suggests the
need to consider alternative therapies discussed later in this chapter.

F ECF FH HE @ @AE C @AE C HE @ D AC FD 02 .4.(1( 4 ( 1. 4. (0 4). 1) 4 ) 1 HFD

1654 PART III Neurological Diseases and Their Treatment

TABLE 100.5 Key Known Antiseizure Medication Mechanisms of Action

High- α2δ-Subunit
Voltage of Voltage-
Activated Activated
Antiseizure Sodium Potassium Enhancing Glutamate Calcium T-Calcium Calcium
Medication Channels Channel GABA Receptor Channels Channels Channels SV2A Comment
Phenobarbital X X X
Primidone X X
Phenytoin X
Methsuximide ? X
Ethosuximide X
Clonazepam X
Carbamazepine X
Valproate X X X
Felbamate X X X X NMDA receptor
antagonism
Gabapentin X
Lamotrigine X X
Topiramate X X X Kainate and AMPA
receptor antag-
onism
Tiagabine X Inhibition of GABA
reuptake
Levetiracetam X
Oxcarbazepine X
Zonisamide X X
Pregabalin X
Lacosamide X Selective enhancing
of slow inactiva-
tion of voltage-
gated sodium
channels
Rufinamide X
Vigabatrin X Irreversible inhi-
bition of GABA
transaminase
Ezogabine/reti- X Enhancing of
gabine transmembrane
potassium
currents
Clobazam X
Perampanel X Noncompetitive
antagonism of
AMPA glutamate
receptors
Eslicarbazepine X
Brivaracetam X
Cannabidiol X Also modulates
intracellular
calcium
Cenobamate X X

AMPA, α-Amino-3-hydroxy-5-methyl-4-isoxazole-propionic acid; GABA, γ-aminobutyric acid; NMDA, N-methyl-D-aspartate; SV2A, synaptic vesicle
protein 2A.

Drug resistance does not always manifest early in the course of epi-
lepsy. It may develop after a variable latency, may be overcome with
the use of a new ASM, and may be punctuated by periods of drug
responsiveness, particularly with the use of new ASMs (Schmidt and
Loscher, 2005a). In the follow-up study by Brodie and colleagues of
1098 patients with newly diagnosed epilepsy, 68% were seizure free
for at least 1 year at the last clinic visit. Upon reviewing the course
of seizure control, 37% achieved early and sustained seizure freedom,
22% had sustained seizure freedom after a delay, 16% had a pattern of
alternating seizure freedom and relapse, while 25% never attained sei-
zure freedom (Brodie et al., 2012). In one group of patients evaluated
for resective epilepsy surgery, the mean latency to failure of the second

F ECF FH HE @ @AE C @AE C HE @ D AC FD 02 .4.(1( 4 ( 1. 4. (0 4). 1) 4 ) 1 HFD


ASM was 9.1 years (range, 0–48), and 26% of patients reported at least
1 year of seizure freedom after the second unprovoked seizure (Berg
et al., 2003).
Pharmacoresistance may have more than one mechanism. One
proposed mechanism is that ASMs do not reach the epileptogenic
zone in sufficient concentration because of multidrug transporters
that transport ASMs out of neurons. A number of findings support
this hypothesis indirectly, but there is no direct evidence of cau-
sality (Schmidt and Loscher, 2009). Another key hypothesis is that
the ASM target (e.g., channels, receptors) is no longer sensitive
to the ASM. Disease severity may play a role and may have com-
mon neurobiological factors with pharmacoresistance (Schmidt
and Loscher, 2009). Another explanation for the loss of an ASM
response is tolerance, which is established with benzodiazepines
but likely occurs with other ASMs as well (Loscher and Schmidt,
2006).
Seizure aggravation may occur with a number of medications
(Chaves and Sander, 2005). This is most common in generalized epi-
lepsy, where ASMs such as carbamazepine, oxcarbazepine, phenytoin,
tiagabine, gabapentin, or vigabatrin may increase the number of sei-
zures or provoke the appearance of new seizure types (myoclonic or
absence seizures) at a therapeutic level. In addition, any ASM may
cause a paradoxical increase in seizures in some patients. This phenom-
enon has been documented with levetiracetam (Nakken et al., 2003).
A number of ASMs have been reported to increase seizure frequency
or severity as a manifestation of toxicity, with ASM serum levels above
“therapeutic” range. This is well documented for phenytoin but may
also occur with other agents. Finally, seizure exacerbation may occur
in the setting of ASM-induced encephalopathy or with sedation. For
example, seizures may be exacerbated with valproate-induced enceph-
alopathy, and sedative ASMs may exacerbate tonic seizures in patients
with Lennox-Gastaut syndrome.
Medication Adverse Effects
It is essential to know the most common and the most serious
adverse effects of ASMs before using them. Such potential adverse
effects should be discussed with the patient before prescribing any
medication. The most common adverse effects are dose dependent
and will predictably occur if titration continues (Toledano and Gil-
Nagel, 2008). Their appearance indicates that the dose should be
reduced. These adverse effects are most likely to occur at the time of
greatest serum concentration following medication intake, in which
case they could be alleviated without dose reduction by splitting the
dose, taking the medication with food, or using an extended-release
preparation. Some dose-dependent adverse effects can be predicted
by the mechanism of action. For example, the most common dose-
dependent adverse effects with medications acting on the sodium
channel (phenytoin, carbamazepine, oxcarbazepine, lamotrigine) are
dizziness, ataxia, and diplopia. Sedation is common and sometimes
unavoidable with benzodiazepines and barbiturates acting on the
GABA receptor. In general, dose-dependent adverse effects are revers-
ible with dose reduction or medication discontinuation. Cognitive
and behavioral adverse effects of ASMs could be considered in the
category of dose-dependent adverse effects, although some individuals
may be predisposed by virtue of genetic or other factors. Barbiturates
and benzodiazepines are best known to affect cognition, but any ASM
can cause impairment of concentration and memory. Among the
new ASMs, topiramate and zonisamide are the most likely to impair
cognition, while lamotrigine is the least likely. Depression may occur
with any ASM, and psychosis is an uncommon adverse effect of many
ASMs but seems more likely associated with topiramate, vigabatrin,
and levetiracetam.
F ECF FH HE @ @AE C @AE C HE @ D AC FD 02 .4.(1( 4 (
CHAPTER 100 Epilepsies 1655
Long-term adverse effects of ASMs are often overlooked but could
be more concerning because they are difficult to reverse or at times
are irreversible. Common examples include weight gain seen with
valproate and pregabalin, and weight loss noted with topiramate
and zonisamide. Barbiturate chronic use may be associated with fro-
zen shoulder and Dupuytren contractures. One of the more recently
appreciated chronic adverse effects of ASMs is reduction in bone min-
eral density. This is most likely to occur with chronic use of enzyme-
inducing ASMs but has also been associated with valproate and other
nonenzyme-inducing agents (Ensrud et al., 2004, 2008; Farhat et al.,
2002). It is unclear whether any ASM is not detrimental to bone den-
sity, and bone mineral density reduction may have different mecha-
nisms with different drugs. It is therefore advisable to monitor bone
density with chronic ASM therapy, and to supplement vitamin D and
calcium. The enzyme-inducing ASMs carbamazepine and phenytoin
have been associated with serological markers of vascular risk, ame-
liorated upon switching to lamotrigine or levetiracetam (Mintzer
et al., 2009). Another long-term adverse effect of ASMs is potential
for teratogenicity (Kluger and Meador, 2008) as well as reduced IQ in
exposed infants. This adverse effect is most pronounced with valproate
(Meador et al., 2009). Valproate in utero exposure also increases risk of
autism (Christensen et al., 2013).
Idiosyncratic adverse effects are not dose dependent but rather
depend on individual patient genetic predisposition. They most
commonly affect organs such as the liver, skin, and blood. Examples
include hepatotoxicity, which may occur with valproate and felbamate;
Stevens-Johnson syndrome, which may occur with several ASMs
including phenytoin, carbamazepine, and lamotrigine; and aplastic
anemia, which may occur with felbamate and (more rarely) carbamaz-
epine. There are some predictors for these serious adverse effects. For
example, valproate hepatotoxicity is more likely in children younger
than 2 years, particularly those with mitochondrial disease (Bjornsson,
2008). Stevens-Johnson syndrome and toxic epidermal necrolysis with
carbamazepine are more likely in individuals of Asian descent who
carry the HLA-B1502 allele. Prior immune disorder and prior cytope-
nia are risk factors for felbamate-associated aplastic anemia (Pellock
et al., 2006). In some instances, genetic testing may identify individuals
at greater risk of idiosyncratic reactions, but there is a need for prog-
ress in the field of epilepsy pharmacogenomics (Kasperaviciute and
Sisodiya, 2009).
Therapeutic Drug Monitoring
Monitoring ASM levels should be used predominantly as an adjunct
to clinical decision making rather than the determinant of ASM dose.
An ASM is typically titrated to the lowest dose known to be effective,
but titration may stop earlier if it is clear that seizure freedom has
already been achieved at a lower dose. Once a clinically effective dose
has been reached, a serum drug level is then obtained for future ref-
erence. Routine drug levels are not necessary, and the level does not
have to be repeated after that unless a breakthrough seizure occurs.
The new level can be compared to the baseline reference level to deter-
mine whether a drop in the level played a role in the seizure. A serum
level can also be helpful to determine how much room there is to
increase the dose if seizures have not yet come under control. ASM
serum levels are valuable to help explain lack of ASM efficacy at what
appears to be a relatively high dose or appearance of adverse effects
at a relatively low dose. One explanation for lack of efficacy is non-
compliance. If seizures have not come under control and the ASM
serum level is 0, it is likely that the patient is not actually taking the
medication. Another reason for a low serum level may be fast metab-
olism, either due to genetic factors or due to enzyme induction by
a concomitant medication. Similarly, serum levels can be useful to
1. 4. (0 4). 1) 4 ) 1 HFD
1656 PART III Neurological Diseases and Their Treatment
explain toxicity at a relatively low dose. Adverse effects may be due to
an elevated serum level as a result of slow metabolism (either due to
genetic factors or to hepatic or renal dysfunction) or pharmacokinetic
interaction.
A therapeutic range is quoted for some ASMs, particularly
classical ASMs such as phenytoin, carbamazepine, and valproate.
Although a useful guide, a value outside the range should never be
the only basis for dosage change. This therapeutic range may be
helpful for patients with infrequent seizures, for whom ascertain-
ment of effective seizure control may take a very long time. For
these patients, aiming for a level in the middle of the range may
be advisable. The classical ASM for which serum levels may be the
most helpful is phenytoin. Phenytoin has nonlinear kinetics, and
its serum level can fluctuate widely with small changes in dose or
bioavailability. When the dose is being titrated in a patient with
difficult-to-control seizures, it may be difficult to predict when
to stop increasing the dose, and the level may have to be checked
intermittently during this process. The usually quoted therapeutic
range is 10–20 µg/mL. If the level is found to be at 20 µg/mL, addi-
tional titration may result in toxicity. Nevertheless, it is ill advised
to reduce the dose for a level above 20 if the patient has good seizure
control and no adverse effects. A therapeutic range is less estab-
lished for many of the new-generation ASMs. Lamotrigine levels
have been better evaluated than other new ASMs. Toxic adverse
effects are likely to occur with a level greater than 20 µg/mL, and
few patients are expected to derive therapeutic benefit beyond this
point (Hirsch et al., 2004).
For two medications that are highly protein bound, protein-free
levels may have added value. Phenytoin and valproate are approxi-
mately 90% protein bound, and the protein-free portion is responsible
for efficacy and toxicity (Levy and Schmidt, 1985). The total level is
usually a good predictor of the free level. However, in instances where
the proportion of protein binding may be altered, the total serum level
is a poor predictor of the protein-free level. These situations include
low-protein states such as malnutrition, hepatic failure, and renal fail-
ure, the combined use of phenytoin and valproate, which compete for
protein binding, and pregnancy. The elderly may also have a greater
protein-free portion. Valproate has saturation kinetics for protein
binding such that there is a greater proportion of free valproate at
higher levels. Protein-free serum levels have greater value in guiding
therapy in the special situations noted.
Discontinuation of Antiseizure Medication Therapy
Owing to the potential adverse effects of ASMs, particularly long-term
adverse effects, discontinuation of ASM therapy should be considered
when it is no longer necessary. Although there are several predictors
of sustained remission on no medications, there is never a guarantee
that seizures will not recur after ASM discontinuation. The decision
to withdraw ASMs is easiest to make when the epileptic syndrome is
known to remit. One example of such a syndrome is BECTS, which is
known to remit at puberty. On the other hand, JME, the most com-
mon idiopathic generalized epilepsy syndrome, is known to have a very
high risk of seizure recurrence; pharmacological therapy is expected to
be lifelong in the majority of patients.
Most patients will not be diagnosed with a specific syndrome whose
course is well known. In a study of 294 children with nonsyndromic
epilepsy followed for more than 10 years, complete remission, defined
as 5 years seizure free and medication free, was achieved by 58% of
children (Berg et al., 2011). Good seizure outcome at 2 years and no
known underlying cause of epilepsy were predictors of remission,
while older age at onset was independently associated with a poorer
chance of complete remission.
F ECF FH HE @ @AE C @AE C HE @ D AC FD
The decision to withdraw medications must balance the potential
consequences of seizure relapse and the potential benefits of eliminat-
ing medication side effects and costs on an individual basis (Shih and
Ochoa, 2009). The risk of seizure recurrence after ASM discontinua-
tion is lower in children than in adults, and the potential psychoso-
cial consequences of recurrent seizures are fewer in children. Hence
ASM withdrawal is considered sooner in children, usually after 1 or
2 years of seizure freedom. In a meta-analysis of ASM discontinua-
tion, the overall risk of relapse after ASM discontinuation was 25%
at 1 year and 29% at 2 years. Relapse was more likely in patients with
remote symptomatic seizures than in patients with idiopathic seizures,
and with adolescent-onset versus childhood-onset epilepsy; an abnor-
mal EEG was associated with a relative risk of 1.45. One study found a
greater predictive value for serial EEGs obtained after initiating ASM
withdrawal (Galimberti et al., 1993). Other factors associated with a
greater risk of recurrence in other studies are focal seizures and longer
than 5 years to attain seizure freedom. Early ASM response predicted
successful ASM withdrawal in children (Shih and Ochoa, 2009). A pro-
spective multicenter randomized study of continued ASM treatment
versus slow withdrawal over 6 months reported that 41% of those in
the withdrawal group relapsed at 2 years, compared to 22% of patients
on continued treatment (Medical Research Council Antiepileptic Drug
Withdrawal Study Group, 1991).
Additional factors predicting recurrence in adults have included
longer duration of active disease, shorter number of years of seizure
remission, abnormal psychiatric examination, presence of hippocam-
pal atrophy, abnormal neurological findings, IQ below 70, increased
number of seizures, focal-onset seizures, and multiple seizure types
(Shih and Ochoa, 2009). Seizure relapse in adults will restrict driving
privileges and may endanger employment, particularly in certain occu-
pations. There is also a concern that seizure control may be difficult to
regain (Schmidt and Loscher, 2005b). Hence, discontinuation of ASMs
in seizure-free adults is usually delayed for at least 4 years and occa-
sionally postponed indefinitely.
The optimal rate of ASM withdrawal after remission has not been
established scientifically. Abrupt discontinuation of ASMs is not
appropriate in this setting. Severe seizures have been reported during
withdrawal of some ASMs, including benzodiazepines, carbamaze-
pine, and oxcarbazepine. It is generally agreed that ASMs in general
and medications associated with tolerance (e.g., benzodiazepines) in
particular should be withdrawn gradually.
SURGICAL THERAPY
Timing
If seizures fail to come under control with medications, surgical ther-
apy is considered. An ILAE task force suggested that failure of two
tolerated, appropriately chosen, and used ASM schedules constitutes
medical failure (Kwan et al., 2010). This was in part based on the
finding that when two ASM monotherapies have failed, the chances
that additional medical therapy will provide complete seizure control
diminish considerably (Kwan and Brodie, 2000a). It is reasonable at
that point to refer the patient to an epilepsy center, particularly if the
underlying epilepsy is surgically remediable and surgical results are
expected to be excellent. Examples include temporal lobe epilepsy with
hippocampal sclerosis and focal epilepsy with underlying focal epilep-
togenic lesion. However, if expected surgical results are less favorable,
it is reasonable to continue medical therapy using different monother-
apies and various combinations (see Fig. 100.19). In a study by Selwa
et al., (2003), continued changes in ASM regimen rendered about 20%
of patients who were not candidates for epilepsy surgery seizure free at
4-year follow-up. It is important to keep in mind that surgical outcome
02 .4.(1( 4 ( 1. 4. (0 4). 1) 4 ) 1 HFD

and chances of seizure freedom after postoperative ASM withdrawal
are better with earlier surgery, so epilepsy surgery should not be
delayed for too long (Simasathien et al., 2013).
Evaluation for epilepsy surgery should probably not be pursued for
patients who are noncompliant with medications—and therefore have
not actually failed medical therapy. Nor should surgery be pursued for
patients with only brief, subjective FAS (i.e., isolated auras) since these
seizures often persist after epilepsy surgery.
Presurgical Evaluation
An extensive presurgical evaluation is necessary before consider-
ing epilepsy surgery and has already been discussed in detail (see
Evaluation and Diagnosis). Its purposes are to (1) localize the epilep-
togenic zone (whose resection is necessary and sufficient to eliminate
seizures), (2) identify incongruent evidence that may indicate the need
for additional tests including invasive EEG, and (3) determine whether
planned surgical resection poses risk to cerebral functions. Essential
elements of the presurgical evaluation include interictal and ictal EEG,
video analysis of recorded seizures, and structural imaging with MRI
to identify an epileptogenic lesion. Interictal PET scanning with FDG
and neuropsychological testing are also commonly part of the pre-
surgical evaluation. In cases where interictal EEG abnormalities are
consistently focal and congruent with a focal unilateral structural epi-
leptogenic lesion, some have advocated skipping ictal EEG recordings,
particularly where resources are limited. Complex scenarios where
there is incongruence in the presurgical data require additional testing
that should include functional imaging with PET, ictal SPECT, MEG,
or even invasive EEG with implanted intracranial electrodes. The deci-
sion to proceed with surgery should balance the predicted benefits of
epilepsy surgery against the predicted risks of functional deficits that
might result from surgery.
Surgical Approaches
Epilepsy surgery can be classified as either curative or palliative.
The aim of curative surgery is to eliminate seizures completely and
potentially produce permanent remission without the need for
ASMs. Palliative surgery is considered only if “curative” surgery is
not viable.
The most common epilepsy localization is mesial temporal, specifi-
cally amygdalohippocampal. The most common surgical approach has
been a temporal lobectomy in which lateral temporal cortex is resected
first, followed by resection of the amygdala and hippocampus. The
resection size is typically different for dominant and nondominant
resection. The lateral resection usually measures around 6 cm from
the temporal pole on the right but is less extensive on the left to reduce
the chance of language deficits. The typical left dominant temporal
lobectomy measures about 4 cm from the temporal pole (Wiebe et al.,
2001). Dominant temporal lobectomies often spare the superior tem-
poral gyrus in an effort to reduce risk of language disturbance, but a
randomized prospective trial did not suggest that this was associated
with benefit (Hermann et al., 1999). Some centers tailor the resection
based on preoperative language functional mapping with electrical
stimulation or fMRI. The hippocampal resection usually measures
about 3 cm. One study suggested that a more complete hippocampal
resection was associated with greater chance of postoperative seizure
freedom (Wyler et al., 1995), but the findings were not confirmed in a
second study (Schramm et al., 2011).
For individuals with clear hippocampal sclerosis, selective amyg-
dalohippocampectomy (SAH) is an alternative approach with less
risk to language functions and equal outcome for seizure control
(Tanriverdi et al., 2008). The original approach advocated by Yasargil
was trans-sylvian (Siegel et al., 1990), but the same operation could
F ECF FH HE @ @AE C @AE C HE @ D AC FD 02 .4.(1( 4 (
CHAPTER 100 Epilepsies 1657
use a transcortical approach through the middle temporal gyrus, or
an inferior temporal approach. No clear difference in surgical out-
come was noted in a study comparing transcortical and trans-sylvian
approaches, but phonemic fluency was significantly improved after
transcortical but not after trans-sylvian SAH (Lutz et al., 2004). The
“Spencer” approach involves resection of the anterior 3 cm of the tem-
poral pole to expose mesial temporal structures for resection (Spencer
et al., 1984). A favorable surgical outcome with respect to seizure con-
trol requires inclusion of the parahippocampal gyrus in the surgical
resection (Siegel et al., 1990). Recently, laser ablation under real-time
MR thermographic guidance has been proposed as an alternative to
open resection in patients with hippocampal sclerosis (Gross et al.,
2018). Its main advantages are decreased surgical morbidity and better
cognitive outcome (Drane et al., 2015). This approach can also be used
for other small deep epileptogenic lesions. Radiofrequency thermo-
ablation has been used less frequently to lesion deep epileptogenic foci.
It can be applied via depth electrodes already implanted for localiza-
tion purposes, but with the disadvantage of less temperature control
(Voges et al., 2018).
Lesionectomy is a suitable surgical approach when there is a well-
defined structural lesion such as benign tumor or cavernous malforma-
tion. In the latter case, the resection must include hemosiderin-stained
tissue surrounding the malformation. There is some evidence to sup-
port the use of intraoperative electrocorticography to guide the extent
of resection (Van Gompel et al., 2009).
Nonlesional neocortical epilepsy usually requires a tailored resec-
tion after the ictal onset zone and cortical functions have been defined
through intracranial recordings, most often using subdural grids.
Hemispherectomy is the preferred surgical approach when the epi-
leptogenic zone is well lateralized but widespread in one hemisphere,
and the hemisphere functions are impaired or expected to become
impaired (Limbrick et al., 2009). Examples of conditions for which
hemispherectomy is often the recommended procedure include
Rasmussen syndrome, Sturge-Weber syndrome, and hemimegalen-
cephaly. The current functional hemispherectomy technique removes
temporal and centroparietal regions, leaving the frontal and occipital
poles with their blood supply but disconnected from the remainder
of the brain. Hemispherectomy provides complete seizure control in
approximately three-quarters of patients and improved seizure control
in the majority of the remainder (Limbrick et al., 2009). The disap-
pearance of seizures will often improve the function of the remain-
ing hemisphere such that cognitive function and behavior are often
improved at follow-up.
If the epileptogenic zone includes eloquent cortex, resection may
not be possible without unacceptable deficits, and the technique of
MST is then considered (Morrell et al., 1989). MST involves discon-
nection of horizontal intracortical fibers while preserving the integrity
of vertical connections. The procedure is based on evidence that the
ictal discharge often spreads along horizontal fibers, while cortical
functions tend to follow a vertical columnar organization. MST on elo-
quent cortex is most often performed in conjunction with resection of
adjacent nonessential cortex. More recently, hippocampal transection
was proposed as a memory-sparing procedure for patients with MTLE
without hippocampal sclerosis, who are at risk of memory decline
from hippocampal resection (Uda et al., 2013).
Corpus callosotomy (CC) is a palliative surgical procedure involv-
ing partial or complete disconnection of the corpus callosum. The
procedure is most often used for drop attacks and is thought to dis-
rupt rapid bilateral seizure spread responsible for sudden loss of con-
sciousness or loss of posture without warning. CC can change seizure
characteristics such that seizures may become focal or patients may
have a warning, giving them time to protect themselves before further
1. 4. (0 4). 1) 4 ) 1 HFD
1658 PART III Neurological Diseases and Their Treatment
seizure progression. In seizure types where bihemispheric synchrony is
required for seizure expression, CC may potentially eliminate clinical
seizure manifestations. Drop attacks are the seizure type most helped
by CC, with benefit in about three-quarters of patients and freedom
from drop attacks in more than a third (Tanriverdi et al., 2009). GTC
seizures are also helped. A two-thirds anterior callosotomy is gener-
ally performed first, with the possibility of extending the callosotomy
later if there is insufficient improvement. CC is not effective for focal
seizures, including FIAS of temporal lobe epilepsy. It is reserved for
patients with severe epilepsy, falls, and injuries. The most appropri-
ate candidates have symptomatic generalized epilepsy, but CC may be
considered for patients with highly refractory GTC seizures in the set-
ting of IGE (Cukiert et al., 2009; Jenssen et al., 2006b).
Surgical Results and Predictors of Surgical Freedom
Epilepsy surgery can be a very effective treatment for patients who are
found to be good candidates after a presurgical evaluation. The best
surgical outcome has been reported in temporal lobe epilepsy sur-
gery. The efficacy of temporal lobe epilepsy surgery was confirmed in
a randomized controlled trial in which 80 patients with temporal lobe
epilepsy were randomly assigned to immediate temporal lobectomy or
additional ASM therapy for 1 year. At 1 year, the difference between
the two groups was highly significant: 58% of patients in the surgical
group and 8% in the medical group were free of seizures impairing
awareness; 38% in the surgical group and 3% in the medical group
were free of all seizures, including auras (Wiebe et al., 2001).
Many other studies have confirmed efficacy of surgery, with gener-
ally better results for MTLE than neocortical epilepsy. In a seven-cen-
ter prospective observational study of resective epilepsy surgery in
patients aged 12 years and older, 339 operated patients (297 mesial
temporal, 42 neocortical) were followed over 2 years (Spencer et al.,
2005). Of these, 66% (223) experienced 2-year remission, not signifi-
cantly different between mesial temporal and neocortical resections
(68% and 50%, respectively). Seizure remission was defined as 2 years
completely seizure free after hospital discharge, with or without auras.
Only absence of GTC seizures and presence of hippocampal atrophy
were significantly and independently associated with remission, and
only in the mesial temporal resection group.
Epilepsy surgery significantly improved quality-of-life score mea-
sures within 6 months after surgery; subsequent changes over time
were sensitive to seizure-free and aura-free status (Spencer et al.,
2007). Other factors reported in some studies as predictors of post-
operative seizure freedom after temporal lobe surgery include discrete
abnormalities (lesions and hippocampal sclerosis), unilateral ictal and
interictal epileptiform discharges, and antecedent febrile convulsions
(not consistently reported). The etiology of head trauma, normal MRI,
and absence of hypometabolism on PET were unfavorable predictors
in some studies. Postoperative seizures and epileptiform activity on
postoperative EEG were also unfavorable predictors. In neocortical
epilepsy, well-circumscribed lesions (e.g., tumor, cavernoma), com-
plete lesion resection, and focal beta or gamma activity at ictal onset
were favorable predictors, while absence of a lesion and generalized
spike-and-wave EEG pattern were unfavorable predictors.
ASMs usually have to be continued for at least a period of time
after successful epilepsy surgery, but the number and dose of ASMs
may be reduced prior to that time. Most commonly, medications are
tapered after 1–2 years of complete seizure freedom. However, there is
a risk of recurrence in approximately one-fifth to one-third of patients
(Kim et al., 2005; Lee et al., 2008; Park et al., 2010b; Rathore et al.,
2011; Schiller et al., 2000; Schmidt et al., 2004). The risk seems lower
in pediatric patients (Lachhwani et al., 2008). Factors associated with
greater likelihood of recurrence included older age at surgery and
F ECF FH HE @ @AE C @AE C HE @ D AC FD
longer duration of epilepsy (Al-Kaylani et al., 2007; Kim et al., 2005;
Lee et al., 2008). Other factors associated with recurrence were ASM
reduction before 10 months (Lee et al., 2008) and normal preoperative
MRI (Schiller et al., 2000).
Late relapse is not a rare event. Relapse after a 2-year remission
occurred in 25% of mesial temporal and 19% of neocortical epilepsy
patients in the seven-center study. Only delay to remission predicted
relapse, and only in mesial temporal epilepsy patients (Spencer et al.,
2005). In a study of anterior temporal lobectomy for temporal lobe
epilepsy in 325 patients followed for a mean of 9.6 years, 55.3% of
patients were seizure free at 2 years, 47.7% at 5 years, and 41% at
10 years (McIntosh et al., 2004). Patients who were seizure free at
2 years postoperatively had a 74% probability of seizure freedom at
10 years after surgery. Long-term postoperative outcome was exam-
ined specifically in 171 patients with MRI-defined hippocampal
sclerosis (Janszky et al., 2005a). Seizure freedom was noted in 80%
at 6 months, 71% at 2 years, 66% at 3 years, and 58% at 5 years.
Predictors of outcome varied at different time points: preoperative
FBTCS and ictal dystonia were negative predictors at 2 years; longer
epilepsy duration and ictal dystonia at 3 years; and only longer epi-
lepsy duration at 5 years.
About a third of patients who are not seizure free immediately after
surgery eventually achieve long-term seizure freedom (Janszky et al.,
2005b). Normal MRI findings and FBTCS preoperatively were unfa-
vorable predictors; rare postoperative seizures and ipsilateral tempo-
ral interictal epileptiform discharges were associated with seizure-free
outcome. Newly administered levetiracetam also showed a significant
positive effect on the postoperative outcome, independent of other
prognostic factors (Janszky et al., 2005b).
Patients who have failed epilepsy surgery can be considered for
reoperation, usually after at least 1–2 years. Reoperation is more likely
to be successful if the initial surgery missed the epileptogenic zone or
epileptogenic lesion, or if the initial resection was incomplete. Between
one- and two-thirds of reoperations result in seizure freedom or
near-seizure freedom (Germano et al., 1994; Gonzalez-Martinez et al.,
2007; Holmes et al., 1999; Ramantani et al., 2013; Salanova et al., 2005;
Siegel et al., 2004). If the first presurgical evaluation was extensive
and the surgery was appropriate based on the findings, reoperation is
unlikely to be successful.
Epilepsy surgery has potential adverse effects that may be expected
in some instances, as well as unexpected complications. A contralateral
upper-quadrant visual-field loss is to be expected after temporal lobec-
tomy, but this is usually not of functional consequence to the patient
(Hughes et al., 1999). Cognitive changes may occur. Dominant tempo-
ral lobe epilepsy surgery carries risks of verbal memory loss in 44% and
reduced naming in 34% of patients (Sherman et al., 2011). Memory
may continue to decline for up to 2 years after surgery (Alpherts et al.,
2006). There is a suggestion of increased risk for memory decline in
patients with larger hippocampal volumes (Baxendale et al., 2008).
On the other hand, memory deficits associated with the function of
the contralateral temporal lobe may improve postoperatively in some
patients with unilateral hippocampal sclerosis (Baxendale et al., 2008).
OTHER THERAPIES
Other therapies are often considered only after medical therapy
fails and when surgical therapy is not possible or has also failed
(Cascino, 2008). However, alternative nonpharmacological ther-
apy should also be considered when patients may be candidates
for surgery, but expectations for complete seizure freedom are low.
These alternatives include dietary therapy, stimulation therapies,
and radiosurgery.
02 .4.(1( 4 ( 1. 4. (0 4). 1) 4 ) 1 HFD

Dietary Therapy
Dietary therapy is a very old treatment of epilepsy, first proposed in
1921 to mimic the effects of fasting by producing ketosis, acidosis, and
dehydration (Sinha and Kossoff, 2005). Interest in a ketogenic diet
decreased with the introduction of many new ASMs in the 1990s but
increased again with the realization that the newer ASMs had only a
modest impact on drug-resistant epilepsy. The ketogenic diet is a very
low-carbohydrate, high-fat, and low-to-adequate protein diet that
includes some restriction of total calories (≈75% of age recommen-
dations). The amount of protein is based on age requirement: carbo-
hydrates are only 5–10 g/day, and the remaining calories come from
fat. The ratio of fat to protein plus carbohydrate ranges from 2:1 to
4:1. The diet is typically initiated with a fast. Although the fasting is
not necessary for therapeutic efficacy, it does provide a faster onset
of action that may help improve compliance. Initiation with fasting
requires hospital admission, which may also be helpful for monitoring
unexpected adverse effects of the diet and reviewing medications for
possible carbohydrate components (Sinha and Kossoff, 2005). Efficacy
of the ketogenic diet in children was confirmed in a randomized con-
trolled but unblinded trial: 38% of children who received the diet had
a greater than 50% reduction versus only 6% of controls, and 7% had
a greater than 90% reduction versus none of the controls (Neal et al.,
2008). A blinded crossover study in children with Lennox-Gastaut syn-
drome did not reach significance, but this may have been due to meth-
odological problems (Freeman et al., 2009). In a prospective study of
the ketogenic diet in 150 children with drug-resistant epilepsy, 3%
were seizure free at 3 months and 7% at 12 months; 27% had a greater
than 90% decrease in seizure frequency at 1 year (Freeman et al., 1998).
A multicenter study similarly showed that 10% of children with highly
refractory epilepsy were seizure free at 1 year, and 40% had a 50% or
greater decrease in seizure frequency (Vining et al., 1998).
The ketogenic diet is a little less effective in adolescents and adults
than in children and is also limited by a higher rate of noncompliance
in adolescents. The onset of action is very fast. In one study the median
time to first improvement was 5 days, with a range of 1–65 days. The
onset of improvement was faster in children who were fasted (5 vs.
14 days), but there was no difference between 1 and 2 days of fast-
ing. Fasting had no effect on long-term efficacy. Some 75% of children
improved within 14 days. Improvement was unlikely if no benefit had
been seen by 2 months, although exceptions did exist (Kossoff et al.,
2008b).
The mechanism of the ketogenic diet is not well understood. Its
benefit may be related to acidosis, ketosis, calorie restriction and
decrease in blood glucose, dehydration, or increase in certain lipids
(Sinha and Kossoff, 2005). Predictors of success included concomitant
use of the ketogenic diet and vagus nerve stimulation (Kossoff et al.,
2007; Morrison et al., 2009). Children receiving phenobarbital in com-
bination with a ketogenic diet were less likely to benefit. Adverse effects
of the ketogenic diet include constipation and worsening of reflux,
both of which can be managed with minor adjustments and stool soft-
eners. Acidosis may occur, mostly at initiation and in association with
acute illness; it can be managed with hydration. Unexplained fatigue
could be helped by supplementation with carnitine. The potential
adverse effect of decreased growth is most likely to occur in the young-
est children. Renal calculi have been reported in 5%–6% of individuals
and may be averted with improved hydration. Hyperlipidemia is of
unknown long-term significance.
Indications for the ketogenic diet include refractory seizures,
regardless of classification. Individuals unable to tolerate ASM therapy
are particularly good candidates for the diet. The ketogenic diet is easi-
est to manage in tube-fed patients or infants receiving formula. A keto-
genic diet is particularly beneficial and could be considered first-line
F ECF FH HE @ @AE C @AE C HE @ D AC FD 02 .4.(1( 4 (
CHAPTER 100 Epilepsies 1659
therapy for children with GLUT deficiency and pyruvate dehydroge-
nase deficiency. Other syndromes where the ketogenic diet has been
reported to be especially beneficial include myoclonic-astatic epilepsy,
tuberous sclerosis, Rett syndrome, Dravet syndrome, and infantile
spasms (Kossoff et al., 2009). Absolute contraindications include mito-
chondrial disorders, pyruvate carboxylase deficiency, and β-oxidation
defects.
Patient compliance with the ketogenic diet has proven difficult,
so other diets have been explored in the treatment of epilepsy. The
modified Atkins diet was created to be more palatable and less restric-
tive than the ketogenic diet. It only restricts carbohydrates (10 g/day
for children and 15 g/day for adults), not protein, fat, or calories. It
is modified from the standard Atkins diet in that the induction phase
limiting carbohydrates is indefinite, and that fat is not only allowed but
also encouraged. Weight loss is not the goal of this diet, but weight loss
has been associated with improvement.
The efficacy of the Atkins diet has been studied prospectively in
children and adults. In one pediatric study, 65% of children had a
greater than 50% reduction in seizures (Kossoff et al., 2006), and in
an adult study, 47% had greater than 50% reduction at 3 months and
33% at 6 months, but 33% discontinued the diet before 3 months. The
modified Atkins diet also seems to work fairly rapidly; median time
to seizure reduction was 2 weeks. In the adult prospective study, a
higher level of ketosis was associated with improvement early on and
weight loss later on (Kossoff et al., 2008c). Another study suggested
consistently strong ketosis was important for maintaining the efficacy
of diet therapy (Kang et al., 2007). The Atkins diet has advantages over
the ketogenic diet in that it is easier to initiate in an outpatient setting
and requires only limited dietician input. It is more tolerable and has
fewer adverse effects than the ketogenic diet. The presence of obesity
in other family members may encourage them to try the diet as well,
thus improving the chances of success for the patient. The modified
Atkins diet has also been proposed as an inexpensive treatment option
in developing countries (Kossoff et al., 2008a).
The suggestion that a lower glucose level played a role in dietary
efficacy prompted a trial of a low glycemic index diet (Muzykewicz
et al., 2009). This diet selectively restricts high glycemic index foods
that produce a substantial postprandial increase in blood glucose and
insulin. The diet allows only low glycemic index carbohydrates, with
an overall carbohydrate intake of 40–60 g/day. There was a greater than
90% improvement in seizure control in about 25% at 3 months, with
another 25% experiencing 50%–90% improvement. There was a cor-
relation between efficacy and blood glucose at 1 month and 12 months
of treatment.
Overall, the advantages of dietary therapy are that it is effective, has
a rapid onset of action, and has different adverse effects than seen with
ASMs. Disadvantages include that dietary therapy may be socially iso-
lating, and compliance is difficult to maintain. The less-restrictive diets
are easier to follow, but they also give more opportunity for cheating
(Muzykewicz et al., 2009).
Vagus Nerve Stimulation
Vagus nerve stimulation (VNS) was first approved in 1997 as the only
device for the adjunctive treatment of refractory focal-onset seizures
in adults and adolescents aged 12 years or older. More recently, VNS
was also FDA approved as adjunctive long-term treatment for chronic
or recurrent depression that has not responded to antidepressants.
Animal data suggested that VNS has acute abortive effects, terminat-
ing seizures that have already started (McLachlan, 1993), and acute
prophylactic effects, reducing seizure frequency and severity during
intermittent stimulation (Takaya et al., 1996). Efficacy in patients
with drug-resistant epilepsy was confirmed in two pivotal randomized
1. 4. (0 4). 1) 4 ) 1 HFD
1660 PART III Neurological Diseases and Their Treatment
blinded controlled studies that demonstrated 24.5%–28% short-term
reduction in seizure frequency (Handforth et al., 1998; The Vagus
Nerve Stimulation Study Group, 1995). The long-term continuation
studies suggested increasing benefit over time, with median seizure
reduction of 34% 3 months after the end of the second double-blind
trial, and 45% by the end of 1 year. In one cohort followed for 12 years,
mean reduction in seizure frequency was 26% after 1 year and 52%
after 12 years of treatment (Uthman et al., 2004). However, seizure
freedom is reported in less than 10% of patients (Ben-Menachem,
2002). Hence, individuals who are excellent candidates for epilepsy
surgery should be advised of the much greater chance of seizure free-
dom with surgical therapy.
Even though there are no randomized trials of VNS in patients with
drug-resistant generalized epilepsy, several studies have supported
VNS efficacy in both idiopathic and symptomatic generalized epi-
lepsy (Ben-Menachem, 2002; Elliott et al., 2011; Kostov et al., 2009;
Ng and Devinsky, 2004); the same is true for VNS efficacy in children.
Thus VNS efficacy does not seem to be specific for a particular form
of epilepsy or particular age. The mechanism of action of VNS for
epilepsy is not totally clear. Approximately 80% of the vagus nerve is
composed of afferent myelinated fibers projecting to the nucleus of
the tractus solitarius, which itself has widespread projections (Ben-
Menachem, 2002). In humans with implanted VNS, imaging of blood
flow with PET showed that VNS increased blood flow in some regions
and decreased blood flow in others. Increased blood flow in the thal-
amus correlated with long-term seizure control (Henry et al., 1999).
In another study using SPECT, acute reduction in amygdala perfusion
and chronic reduction in hippocampal perfusion correlated with clin-
ical efficacy (Van Laere et al., 2002).
The VNS device consists of a generator, usually implanted over
the chest under the left clavicle, and electrodes that are placed around
the left vagus nerve and tunneled under the skin to connect with the
generator. VNS implantation is usually an outpatient surgical proce-
dure with rare complications, the most important of which is infec-
tion. Asystole (with full recovery) has been reported during routine
intraoperative lead testing, approximately once for every 1000 implan-
tations. VNS is usually programmed to stimulate for 30 seconds, alter-
nating with 5 minutes of rest, but the duration of stimulation time
on and stimulation time off can be changed. Some patients seem to
derive greater benefit from “rapid-cycle” stimulation, with 7 seconds
of stimulation alternating with 12 seconds of rest. The first parameter
to be titrated is current intensity, which is increased by 0.25 mA steps
as needed to achieve benefit. Other output current parameters that can
be programmed are frequency and pulse width. The optimal stimu-
lation parameters have not been well defined and may vary between
individuals. In addition to the recurring output current cycles, a single
on-demand stimulation train can be programmed separately to abort
seizures with the use of a magnet. On-demand stimulation is particu-
larly helpful to abort or attenuate seizures in individuals who have an
aura (Morris, 2003). However, the clinical efficacy of on-demand stim-
ulation is difficult to confirm with rigorous investigation. Newer VNS
models also offer responsive stimulation upon detection of sudden
increase in heart rate. This is based on the observation that ictal tachy-
cardia is an early sign observed in the majority of patients (Eggleston
et al., 2014; Hirsch et al., 2015).
The most common VNS adverse effect is hoarseness, which occurs
during stimulation cycles. This is to be expected in most patients, but it
does improve over time. The same is true of other stimulation-related
adverse effects of coughing, throat pain, dyspnea, and paresthesias
(Ben-Menachem, 2002). VNS may also exacerbate obstructive sleep
apnea, so it is important to diagnose and treat sleep apnea before initi-
ating VNS therapy (Malow et al., 2000a; Marzec et al., 2003).
F ECF FH HE @ @AE C @AE C HE @ D AC FD
The main advantages of VNS are that it does not have the ASM
adverse effects of sleepiness, tiredness, dizziness, or cognitive dys-
function. It may improve mood and promote alertness, and it gives
patients and families a sense of control with the use of on-demand
stimulation to abort seizures. In addition, compliance does not
require patient effort. However, it does require surgical implanta-
tion, and the battery has to be changed every 3–10 years depend-
ing on the stimulation parameters. It is difficult to predict who will
benefit from this therapy, but the best candidates are patients who
are not good candidates for epilepsy surgery, have frequent seizures,
and have a consistent aura or a slow seizure progression, so that on-
demand stimulation could be used to abort seizures. VNS has also
become frequently used in symptomatic generalized epilepsy before
proceeding with CC (Nei et al., 2006; Rosenfeld and Roberts, 2009;
You et al., 2008).
Other Stimulation Therapies
Trigeminal nerve stimulation, which had an antiepileptic effect in a
rodent model of epilepsy, can be delivered noninvasively in humans
and is being investigated as an alternative stimulation modality. In a
small open-label pilot study, bilateral stimulation of the ophthalmic
branch produced a mean reduction in seizure frequency of 59% at 12
months (DeGiorgio et al., 2009). A larger blinded randomized con-
trolled trial in 50 subjects with focal-onset seizures showed a reduc-
tion in seizure frequency as measured by the response ratio, but there
was no significant difference between groups in the 50% responder
rates (30.2% for the treatment group and 21.1% for the active control
group) (DeGiorgio et al., 2013). Of note is that the active treatment
group showed a significant improvement in responder rate over the
treatment period, from 17.8% at 6 weeks to 40.5% at 18 weeks.
Repetitive transcranial magnetic stimulation (rTMS), a noninvasive
cortical stimulation method, was also investigated as a treatment for
drug-resistant epilepsy, with variable results:L rTMS modulates cortical
excitability, with high-frequency rTMS enhancing and low-frequency
rTMS decreasing cortical excitability in most individuals (Gangitano
et al., 2002). Most studies used daily rTMS sessions for about 1 week,
then evaluated efficacy 2–4 weeks later. A meta-analysis deduced that
low-frequency rTMS is moderately beneficial, with more improvement
in subjects who have cortical dysplasia or neocortical epilepsy, proba-
bly because of greater and more precise access of rTMS therapy to the
more superficial seizure foci (Hsu et al., 2011).
Various brain targets have been explored for stimulation, includ-
ing cortical and subcortical targets. Scheduled open-loop stimulation
to various cortical and subcortical structures, including the thalamus,
subthalamic nucleus, cerebellum, and hippocampus, demonstrated
variable success (Jobst et al., 2010a). Bilateral stimulation of the ante-
rior nucleus of the thalamus was proven effective in a multicenter
double-blind randomized trial (Fisher et al., 2010). Median seizure
reduction was 14.5% in the control group and 40.4% in the stimu-
lated group during the blinded phase, and there was a 56% median
reduction in seizure frequency by 2 years. However, participants in the
stimulated group were more likely to report depression or memory
problems as adverse events. Bilateral deep brain stimulation of the
anterior nucleus of the thalamus for epilepsy was first approved in
Europe in 2010, then by the US FDA in 2018 (Salanova, 2018). The
FDA approval is for adjunctive therapy in individuals 18 years of age
or older diagnosed with focal epilepsy “refractory to three or more
anti-epileptic medications.”
Responsive closed-loop stimulation delivers a stimulus to the pre-
sumed seizure onset zone in response to seizure detection (Jobst et al.,
2010a). The concept is based on evidence that brief stimulation can
terminate seizure activity if delivered early after seizure onset. The
02 .4.(1( 4 ( 1. 4. (0 4). 1) 4 ) 1 HFD

generator is implanted in the skull and connected to either depth or
subdural strip electrodes to deliver stimulation directly to one or two
seizure onset zones. The responsive stimulator device was found to
be effective in a pivotal randomized double-blind, sham stimulation
controlled trial in patients with drug-resistant focal-onset seizures. The
reduction in seizure frequency over the blinded evaluation period was
37.9% in the treatment group compared to 17.3% in the sham stim-
ulation group (Morrell, 2011). In the open-label extension, median
percent reduction in seizures was 44% at 1 year and 53% at 2 years,
suggesting progressive improvement with time (Heck et al., 2014). The
device was approved by the FDA as an adjunctive therapy in individ-
uals 18 years or older with focal-onset seizures who have undergone
diagnostic testing that localized no more than two epileptogenic foci,
are refractory to two or more antiseizure AEMs, and currently have
frequent and disabling seizures. Responsive stimulation is a suitable
treatment option for patients with bilateral independent seizure foci or
with an epileptogenic zone in eloquent cortex not suitable for surgical
resection.
In general, the advantages of stimulation include that it is reversible
and adjustable, unlike resective surgery. However, the optimal stimu-
lation parameters are not generally well defined, and, to date, stimu-
lation therapies have been predominantly palliative. The decision to
pursue stimulation therapy has to balance risks and benefits in com-
parison with other available therapies.
Radiosurgery
Radiosurgery uses a stereotactic frame to immobilize the head while
radiation beams are precisely directed from different angles to a target.
The method delivers radiation to the target with a steep gradient so that
regions within a few millimeters of the target receive a substantially
reduced radiation dose (Romanelli and Anschel, 2006). Radiosurgery
was initially used for treatment of brain tumors and AVMs not readily
accessible to standard surgery, with beneficial effect on seizure control.
It has also been used successfully for hypothalamic hamartoma. More
recently, radiosurgery was explored for MTLE. A prospective study in
three European centers reported 65% of patients seizure free at 2 years
after radiosurgery. Five patients had transient side effects, including
depression, headache, nausea, vomiting, and imbalance. No perma-
nent neurological deficit was reported, except nine visual-field defi-
cits. However, seizure freedom was delayed for most patients, the main
improvement occurring between 12 and 18 months; some patients
only became seizure free after 2 years post treatment.
A US multicenter prospective study randomized patients to high- or
low-dose radiosurgery. At 36 months of follow-up, 76.9% of patients
randomized to the high dose and 58.8% randomized to the low dose
were free of seizures for the prior 12 months (Barbaro et al., 2009).
Seizure remission correlated with appearance of vasogenic edema
demonstrated on serial imaging after approximately 9–12 months
(Chang et al., 2010). The degree of radiation-induced local vascular
insult and neuronal loss was dose dependent and predicted long-term
seizure remission (Chang et al., 2010). Neuropsychological testing
showed no definite change in cognitive measures from baseline at 2
years after radiosurgery (Quigg et al., 2011). A single-blinded multi-
center controlled trial randomized 58 adult subjects with drug-resis-
tant unilateral MTLE to either stereotactic radiosurgery or standard
temporal lobectomy (Barbaro et al., 2018). Seizure remission was
achieved in 52% of the radiosurgery and 78% of the temporal lobec-
tomy patients.
Radiosurgery may have a place in the treatment of drug-resistant
mesial temporal epilepsy for patients who are opposed to or at greater
risk for complications with standard epilepsy surgery. However, the
long-term risk/benefit ratio of radiosurgery needs better definition.
F ECF FH HE @ @AE C @AE C HE @ D AC FD 02 .4.(1( 4 (
CHAPTER 100 Epilepsies 1661
QUALITY-OF-CARE STANDARDS IN THE
MANAGEMENT OF EPILEPSY
The American Academy of Neurology developed standardized quality
measures for epilepsy care (Fountain et al., 2011) that have since been
updated based on new evidence and performance gaps (Fountain et al.,
2015; Patel et al., 2018). The latest quality measurement set includes
the following:
year
patients with intractable epilepsy
-
ment
The measures were primarily developed for quality improvement
projects. Providers were encouraged to identify measures most mean-
ingful for their patients and to implement these measures to drive
improvement in practice (Patel et al., 2018).
SEIZURE CLUSTERS AND STATUS EPILEPTICUS
Seizure clusters, also called acute repetitive seizures and serial seizures, are
closely grouped seizures representing an increase in seizure frequency
compared to baseline, usually occurring over the span of minutes to a
couple days. Seizure clusters may include any type of seizure and may
vary in severity, but by definition there is complete recovery in between
seizures. Seizure clusters are more common in patients with drug-
resistant epilepsy, particularly those with remote symptomatic epilepsy
and extratemporal epilepsy. Patients with seizure clusters are more likely
to have a history of status epilepticus. Seizure clusters themselves may or
may not progress to prolonged seizures or even status epilepticus. Such
progression may be predictable for individual patients, based on their
seizure history. This may determine the most appropriate treatment for
seizure clusters. Mild clusters can be treated with oral doses of benzo-
diazepines. However more severe clusters, particularly those known to
progress to severe prolonged seizures or status epilepticus, may require
other routes of administration. Rectal diazepam was the only FDA-
approved treatment for out-of-hospital administration by nonmedical
caregivers (Cereghino et al., 1998), until intranasal midazolam spray was
approved in 2019 (Detyniecki et al., 2019). Buccal midazolam is in wide
clinical use in Europe and various countries (Nakken and Lossius, 2011),
but has not been approved in the United States. Intranasal diazepam was
approved by the US FDA in 2020. The efficacy of intramuscular diaze-
pam delivered by autoinjector was demonstrated in a blinded controlled
trial (Abou-Khalil et al., 2013), but this did not lead to FDA approval or
marketing. Other approaches that were evaluated include buccal diaze-
pam and staccato midazolam.
Status epilepticus was previously broadly defined as seizure activ-
ity that continues for 30 minutes, or recurrent seizures without
recovery between attacks. The 30-minute duration has been the
subject of debate, since it may delay aggressive therapy, particularly
when prolonged duration can be predicted in the absence of therapy.
Experimental evidence suggests that irreversible neuronal injury may
start after 20–30 minutes of generalized convulsive status epilepticus
(GCSE) (Fujikawa, 1996; Meldrum and Brierley, 1973), so every effort
has to be made to stop seizure activity prior to that. A large body of
evidence suggests that the bilateral tonic-clonic phase if focal or gener-
alized onset seizures does not last longer than 2 minutes (Jenssen et al.,
2006a; Theodore et al., 1994) except when it evolves into status epilep-
ticus. As a result, it has been suggested that vigorous therapy for status
1. 4. (0 4). 1) 4 ) 1 HFD
1662 PART III Neurological Diseases and Their Treatment
epilepticus be initiated after 5 minutes of bilateral tonic-clonic activity
(Lowenstein et al., 1999). There is also evidence that FIAS that last lon-
ger than 10 minutes will likely evolve into status epilepticus (Jenssen
et al., 2006a). Based on the above, the ILAE defined status epilepticus
as a condition “resulting from the failure of the mechanisms respon-
sible for seizure termination or from the initiation of mechanisms
which lead to abnormally prolonged seizures (after time point t1),
and can have long-term consequences (after time point t2)” (Trinka
et al., 2015). The time point t1 for status epilepticus was 5 minutes for
bilateral tonic-clonic seizures, 10 minutes for focal seizures, and 10–15
minutes for absence seizures.
Any seizure type may evolve into status epilepticus. Status epilep-
ticus may be classified based on the seizure type it evolves from. The
most dangerous type of status epilepticus, GCSE, may evolve from a
primary GTC seizure or more often from a FBTCS. One form of sta-
tus epilepticus with a low likelihood of irreversible neuronal injury is
generalized absence status epilepticus, which evolves from generalized
absence seizures.
The most recent ILAE classification of status epilepticus (see Box
100.3) includes the two major categories of status epilepticus with prom-
inent motor symptoms and status epilepticus without prominent motor
symptoms, which is synonymous with nonconvulsive status epilepticus.
Status epilepticus with prominent motor symptoms includes convulsive
status epilepticus, which has been made synonymous with tonic-clonic sta-
tus epilepticus, myoclonic status epilepticus (with coma or without coma),
focal motor status epilepticus, tonic status epilepticus, and hyperkinetic status
epilepticus. Convulsive status epilepticus can be generalized, focal evolv-
ing to bilateral tonic-clonic, or unknown whether focal or generalized in
onset. Nonconvulsive status epilepticus can be with coma or without coma.
Nonconvulsive status epilepticus without coma can be generalized as in
absence status epilepticus, or focal, with or without impairment of con-
sciousness. Focal nonconvulsive status epilepticus without impairment of
consciousness is often referred to as aura continua.
Nonconvulsive status epilepticus without coma is generally consid-
ered a less serious medical emergency than convulsive status epilepti-
cus. However, nonconvulsive status epilepticus with coma, including
what was previously referred to as subtle convulsive status epilepti-
cus is extremely serious, difficult to treat, and has a poor prognosis.
Nonconvulsive status epilepticus with coma may evolve from convul-
sive status epilepticus in which treatment has been delayed or has been
ineffective. It may also be seen without prior overt convulsive status
epilepticus when there has been a serious brain insult such as severe
traumatic brain injury, hypoxic injury, or massive strokes. In gener-
alized absence status epilepticus, the patient usually appears awake
but may be either unresponsive to verbal stimuli or may have slowed
or inappropriate responses. In focal nonconvulsive status epilepticus
with impaired consciousness, there is a wide spectrum of impairment
of consciousness, responsiveness, or behavior. There may be consider-
able cyclical fluctuations in the clinical manifestations.
The definitive diagnosis of nonconvulsive status epilepticus
requires confirmation with EEG. Even convulsive status epilepticus
has to be differentiated from psychogenic status epilepticus or pseudo-
status epilepticus, which may require video-EEG for definitive diagno-
sis. However, the EEG patterns of status epilepticus, particularly non-
convulsive status epilepticus, are not totally specific (Kaplan, 2007).
When the EEG pattern is not specific, the definitive diagnosis also
requires observation of a definite clinical response to treatment, or at
least restoration of normal EEG activity that was previously absent. In
convulsive status epilepticus, a sequence of EEG patterns was reported,
starting with discrete recurrent seizures separated by interictal slow
activity, merging seizures with waxing and waning ictal discharge,
continuous ictal discharge, continuous ictal discharge punctuated by
F ECF FH HE @ @AE C @AE C HE @ D AC FD
generalized attenuation, and then periodic discharges on a relatively
flat background (Treiman et al., 1990). This sequence was proven in an
animal model of status epilepticus, and also suspected in human status
epilepticus based on EEGs of patients studied at different stages of the
condition. A similar sequence may also occur in focal status epilepti-
cus, with asymmetrical or focal EEG features. The last pattern in the
sequence, periodic discharges on a relatively flat background, is some-
what problematic because periodic discharges are not specific.
The incidence of status epilepticus is likely underestimated by pub-
lished studies. The highest overall incidence, 41 cases per 100,000 per
year, was found in the prospective population-based study of status
epilepticus in Richmond, Virginia (DeLorenzo et al., 1996). The inci-
dence of status epilepticus is elevated early in life, decreases after that,
then increases in the elderly, with up to 98.9 annual cases per 100,000
persons in that age group (Vignatelli et al., 2003).
The etiology of status epilepticus is very dependent on age. The
most common cause in children is febrile status epilepticus, account-
ing for more than half of cases (Rosenow et al., 2007). In adults, status
epilepticus is much more often due to acute cerebrovascular accidents,
hypoxia, metabolic causes, and low ASM levels (Rosenow et al., 2007).
It is important to recognize that the majority of patients in status epi-
lepticus do not have a history of epilepsy.
Status epilepticus is a neurological emergency that requires prompt
intervention. The goal of therapy is to stop seizure activity in the brain
before neuronal injury has started. In addition, delay in initiating ther-
apy is associated with resistance to treatment (Treiman et al., 1998),
probably due to alteration in the functional properties of GABAA
receptors (Goodkin et al., 2008; Jones et al., 2002; Macdonald and
Kapur, 1999). Since convulsive status epilepticus is the most serious
form, its treatment will be discussed first. Treatment of status epilepti-
cus may have to start before arrival in the emergency room. Individuals
known to have recurrent status epilepticus may respond to rectal diaz-
epam administered by a parent or care partner. There is also evidence
to support the use of buccal midazolam, nasal midazolam, and nasal
lorazepam (Arya et al., 2011). Even when prehospital treatment was
not effective and status epilepticus had persisted upon arrival in the
emergency room, there was evidence that prehospital treatment with
rectal diazepam was associated with a shorter duration of status after
arrival to the emergency department (Chin et al., 2008). For individ-
uals not prepared for home treatment of status epilepticus, the use of
2 mg of IV lorazepam by paramedics was associated with termination
of status before arrival to the emergency room in 59.1% of individu-
als, compared to 21.1% treated with placebo (Alldredge et al., 2001).
Intramuscular autoinjector administration of midazolam by paramed-
ics was found at least as safe and effective as IV lorazepam for prehospi-
tal seizure cessation (Silbergleit et al., 2012). Intramuscular midazolam
could be administered faster than IV lorazepam. At arrival in the
emergency department, 73.4% of 448 subjects in the intramuscular
midazo-lam group were free of seizure activity, as compared with 63%
of 445 subjects in the IV-lorazepam group.
Upon arrival in the emergency room, treatment of convulsive sta-
tus epilepticus begins with basic life support measures, specifically
attention to airway, breathing, and circulation (Treiman, 2007). Blood
samples should be drawn for hematological and serum chemistry val-
ues, as well as ASM levels for patients who are already taking these
medications. If blood glucose is low or cannot be measured rapidly,
IV glucose should be administered, in conjunction with IV thiamine if
there is a concern for malnutrition. Based on the landmark Veterans
Affairs Status Epilepticus Cooperative Study, lorazepam 0.1 mg/
kg was the most effective agent, terminating overt convulsive status
epilepticus within 20 minutes in 64.9% of patients (Treiman et al.,
1998). It was much less effective in subtle convulsive status epilepticus
02 .4.(1( 4 ( 1. 4. (0 4). 1) 4 ) 1 HFD

(nonconvulsive status epilepticus with coma), terminating it within 20
minutes in only 17.9% of patients with a verified diagnosis. Lorazepam
was superior to phenytoin alone, which controlled overt convulsive
status within 20 minutes in only 43.6% of patients with verified diag-
nosis. If the first treatment failed to control status epilepticus, the
chances of control with the second treatment were minimal and mor-
tality was twice as high (Treiman et al., 1998).
A standard treatment protocol is necessary for the hospital treat-
ment of convulsive status epilepticus. The American Epilepsy Society
issued an evidence-based guideline in 2016 (Glauser et al., 2016). It
recommended initiating treatment in adults with either intramus-
cular midazolam, IV lorazepam, IV diazepam, or IV phenobarbital
for a bilateral convulsive seizure that has lasted at least 5 minutes.
Lorazepam is usually the first agent used for terminating status. If lora-
zepam is successful in stopping seizure activity, the decision to add
another agent depends on the underlying etiology. Lorazepam’s dura-
tion of action is approximately 12–24 hours. If the etiology is reversible
(e.g., status epilepticus due to metabolic or toxic factors), lorazepam
may be the only treatment necessary. Intravenous diazepam is not rec-
ommended as the only treatment because of its rapid redistribution in
adipose tissue, markedly shortening its duration of action following IV
administration. Another longer-acting ASM is needed if the underly-
ing etiology is not rapidly reversible.
If convulsive status epilepticus is not controlled after lorazepam 0.1
mg/kg, a second-line therapy should be initiated within 20 minutes of
seizure onset. Choices include IV infusion of 20 mg/kg of fosphenytoin
(no faster than 150 mg/min) or 40 mg/kg of valproic acid, or 60 mg/
kg of levetiracetam (Kapur et al 2019). Lacosamide is a more recently
introduced option, supported with retrospective studies. If convulsive
status epilepticus is still not responsive after 40 minutes, endotracheal
intubation is necessary at this point, followed by general anesthesia
using midazo-lam, propofol, or pentobarbital. The purpose of general
anesthesia is to control electrical status epilepticus in the brain. EEG
monitoring is necessary at this point because electrical status epilep-
ticus may continue after motor activity stops. If the EEG continues to
show ictal activity, the anesthesia has to be deepened to a burst-sup-
pression pattern and even to complete suppression in some cases.
Patients who require general anesthesia to control status epilepticus
will generally require long-term maintenance with ASMs. Serum levels
of ASMs must be in a high therapeutic zone prior to discontinuation of
F ECF FH HE @ @AE C @AE C HE @ D AC FD 02 .4.(1( 4 (
CHAPTER 100 Epilepsies 1663
general anesthesia. If seizure activity recurs upon withdrawal of general
anesthesia, a number of agents have been used successfully. ASMs with
IV formulation are used more often (Agarwal et al., 2007; Aiguabella
et al., 2011; Albers et al., 2011; Alvarez et al., 2011; Chen et al., 2011;
Goodwin et al., 2011; Kellinghaus et al., 2011; Koubeissi et al., 2011;
Misra et al., 2006; Moddel et al., 2009), but high-dose topiramate and
felbamate have also been advocated as treatment options (Wasterlain
and Chen, 2008). During the treatment of refractory status epilepti-
cus with ASMs, the cause of status has to be investigated and treated
appropriately (Shorvon, 2011). Immunotherapy may be indicated if an
autoimmune therapy is suspected.
The treatment of other forms of status epilepticus may be less
aggressive, depending on the type of status encountered. In noncon-
vulsive status epilepticus with impaired consciousness, or focal motor
status epilepticus, aggressive treatment should avoid depressing level
of consciousness. It is still recommended to start with one of the ben-
zodiazepines listed above, followed by second-line agent IV agents,
including fosphenytoin, valproate, levetiracetam, and lacosamide.
The use of general anesthesia should be avoided if at all possible,
because of associated increased risk of infection and death (Sutter
et al., 2014).
Generalized absence status epilepticus may respond to IV loraze-
pam but may require additional IV valproate for control. Generalized
myoclonic status epilepticus can be treated with IV lorazepam, val-
proate, or IV levetiracetam. Levetiracetam is the only ASM approved
by the FDA for treatment of myoclonic seizures based on class I clinical
trial evidence.
Outcome of status epilepticus depends on the underlying cause,
patient age, duration and severity of status epilepticus, and rapidity of
therapy initiation. In the Richmond status epilepticus study, mortal-
ity was 3% for children and 26% for adults (DeLorenzo et al., 1996).
Status epilepticus of less than 1-hour duration had a 2.7% mortality
rate after 1 month, compared to 32% for status epilepticus persisting
longer than 1 hour (Towne et al., 1994). Patient age and depth of coma
at presentation were the strongest predictors of outcome (Neligan and
Shorvon, 2011). However, depth of coma is related to duration of sta-
tus (including delay in initiating therapy) and underlying illness.
The complete reference list is available online at https://expertconsult.
inkling.com/.
1. 4. (0 4). 1) 4 ) 1 HFD